| Abstract [eng] |
Fuel price dynamics have a strong impact on the performance of the transport, logistics, and manufacturing sectors. Since changes in retail gasoline and diesel prices directly affect household spending and business operating costs, analyzing and forecasting these price movements is important for shaping economic policy and helping market participants make informed decisions. The object of this study is the dynamics of retail gasoline and diesel prices in Lithuania. The main aim is to assess the key factors influencing retail fuel prices, evaluate the impact of external shocks, and identify the most accurate forecasting models for gasoline and diesel prices. The first part of the study reviews scientific literature on the concept of fuel, the classification of petroleum products, the main factors affecting fuel price changes, and the econometric forecasting methods used in similar studies. The literature review showed that fuel price dynamics are complex and are mainly influenced by global crude oil prices, domestic economic indicators, and unexpected external shocks. The second part describes the data used in the research, the data preparation process, and the applied modeling methods. Monthly data on retail gasoline and diesel prices in Lithuania, macroeconomic indicators, and dummy variables representing external shocks were used to achieve the research objective. Both univariate and multivariate models were applied, while relationships between dependent and independent variables were identified using the Granger causality test. Model accuracy was evaluated using RMSE and MAPE error measures. The results showed that Prophet-based models provided the most accurate forecasts for both retail gasoline and diesel prices. Gasoline price dynamics were mainly affected by changes in Brent crude oil prices and Lithuania’s unemployment rate, while diesel price movements were influenced by Brent crude oil prices and the Harmonised Index of Consumer Prices in the liquid fuel market. The study also found that the COVID-19 pandemic had a statistically significant effect on retail diesel price dynamics. Forecasts for 2026 showed that Prophet models predicted both fuel price trends relatively accurately at the beginning of the year, but forecasting accuracy decreased significantly in the spring due to rising geopolitical tensions in the Middle East. These results suggest that while Prophet models can effectively capture long-term fuel market trends under normal conditions, their forecasting accuracy becomes much weaker during periods of sudden structural change caused by major geopolitical or economic shocks. |
